[SkLearn TrainTestSplit OneHot Behaviour] #JupyterNotebook #CodeSnippet #@todo: BUG: `stratify=data_holder.Y_data` instead of `stratify=data_holder.Y_data` #@bug: ¿strange behaviour? If stratify is one_hot then changes w.r.t string or int encoded #@bug: See SkLearn_TrainTestSplit_OneHot_Behaviour.ipynb #Others

SkLearn_TrainTestSplit_OneHot_Behaviour.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sklearn Train/Test split random test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from keras.utils import np_utils\n",
    "\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "ids = np.arange(20)\n",
    "ids_2 = np.random.normal(size=(20,))\n",
    "\n",
    "classes = ['label1', 'label2', 'label3']\n",
    "\n",
    "labels = np.random.choice(classes, size=20)\n",
    "\n",
    "label_encoder = LabelEncoder()\n",
    "\n",
    "encoded_labels = label_encoder.fit_transform(labels)\n",
    "\n",
    "one_hot_labels = np_utils.to_categorical(encoded_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Labels: ['label3' 'label3' 'label2' 'label3' 'label2' 'label2' 'label3' 'label3'\n",
      " 'label3' 'label1' 'label2' 'label3' 'label1' 'label1' 'label2' 'label1'\n",
      " 'label2' 'label3' 'label1' 'label2']\n",
      "Encoded labels: [2 2 1 2 1 1 2 2 2 0 1 2 0 0 1 0 1 2 0 1]\n",
      "OneHot labels: \n",
      "[[0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 1. 0.]\n",
      " [0. 0. 1.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [1. 0. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 0. 1.]\n",
      " [1. 0. 0.]\n",
      " [1. 0. 0.]\n",
      " [0. 1. 0.]\n",
      " [1. 0. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 0. 1.]\n",
      " [1. 0. 0.]\n",
      " [0. 1. 0.]]\n"
     ]
    }
   ],
   "source": [
    "print('Labels: {}'.format(labels))\n",
    "print('Encoded labels: {}'.format(encoded_labels))\n",
    "print('OneHot labels: \\n{}'.format(one_hot_labels))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([ 6,  5,  7, 11, 12,  3, 18,  4, 15, 10,  8,  1, 14, 16]),\n",
       " array([ 2, 13,  0, 17,  9, 19]))"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tr_ids, te_ids, tr_labels, te_labels = train_test_split(ids, labels, stratify=labels, test_size=0.3, random_state=2019)\n",
    "tr_ids, te_ids"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([ 6,  5,  7, 11, 12,  3, 18,  4, 15, 10,  8,  1, 14, 16]),\n",
       " array([ 2, 13,  0, 17,  9, 19]))"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tr_ids, te_ids, tr_labels, te_labels = train_test_split(ids, encoded_labels, stratify=encoded_labels, test_size=0.3, random_state=2019)\n",
    "tr_ids, te_ids"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([19,  2, 15, 18,  5,  7, 11,  1, 17, 13, 10,  4,  8,  6]),\n",
       " array([ 3,  9, 12,  0, 16, 14]))"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tr_ids, te_ids, tr_labels, te_labels = train_test_split(ids, one_hot_labels, stratify=one_hot_labels, test_size=0.3, random_state=2019)\n",
    "tr_ids, te_ids"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([[0., 0., 1.],\n",
       "        [1., 0., 0.],\n",
       "        [1., 0., 0.],\n",
       "        [0., 0., 1.],\n",
       "        [0., 1., 0.],\n",
       "        [0., 1., 0.]], dtype=float32),\n",
       " array(['label3', 'label1', 'label1', 'label3', 'label2', 'label2'],\n",
       "       dtype='<U6'))"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "te_labels, label_encoder.classes_[np.argmax(te_labels, 1)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}